As display devices of various sizes and increased resolution have been developed and the demand for them have increased, the ability for a graphics processing system to resize and resample source images and create destination images to take advantage of the various sized and higher resolution displays is a desirable operation. In an electronic display system, color at each pixel is represented by a set of color components, and each color component is represented by a sample value. Color components such as red, green, blue (RGB) or other representations such as YCbCr are well known in the art. Whichever representation is chosen, each color component can be interpreted as a two dimensional array of samples, so three such arrays can represent images on display systems. Conceptually, resampling can be viewed as a spatial process, working on discrete input samples, represented by pixels of the source image arranged in a two-dimensional bitmap. The output samples of the destination image are spatially located at fractional sample positions within the input sample grid. Various interpolation and modeling methods are used to construct transition models between samples of the source image from which additional graphics data is produced during the resampling operation.
The additional graphics data is then used to produce larger or higher resolution destination graphics images. However, the resulting destination image must retain an acceptable image quality with respect to the source image. That is, the destination image should appear to retain at least a similar visual qualities of the source image, such as having nearly the same color balance, contrast, and brightness as the original source image. Otherwise, rather than accurately reproducing a larger or higher resolution graphics image of the source image, the resampling operation will compromise image quality by introducing image distortion. To this end, various resampling algorithms have been developed in order to create high quality destination graphics images.
With many conventional resampling algorithms, a transition model between input samples along each axis is constructed to provide output sample values. Generally good results can be obtained with separable processing along each axis for graphics images because image feature cross-sections have the same characteristics when viewed at any angle within the image plane, only at different effective sample rates. That is, the resampling operation is first performed on the pixels along a first axis of the source image followed by resampling of the pixels along the other axis. The output sample values of the resampling operation along the first axis is then used in generating the transition models from which output samples are calculated. The transition models between the input samples are constructed such that the output samples interpolated from the transition model create a destination image that closely resembles the original or source image. The transition models are typically continuous so that an output sample can be generated at any position between the input samples.
A typical axis separable resampling circuit performs the resampling operation for the source image along one axis in its entirety before beginning the resampling operation along the second axis. In operation, input samples of a source image are scanned in using a row-by-row access and provided to a resampling circuit. The output sample values calculated by the resampling circuit are written to a memory from which the output sample values are accessed when performing the resampling operation along the other axis. The resulting output samples are then written to a display memory from which the data for a destination image are read for rendering. Although conceptually the simplest, it is not a very efficient manner in which to perform resampling. One reason is related to the manner in which a memory device is accessed. The x- and y-axes of the source image correspond to the row and columns of a memory device. Accessing the data for y-axis resampling is ordinarily not efficient because new row and column addresses must be provided for reading or writing of data. In contrast, accessing input samples along a row of the source image, and consequently a row of memory, can take advantage of page or burst access features of the memory device. These features typically require providing the column address for access along a row, or in some instances, simply strobing a single command signal is all that is necessary to increment the column address once a row address is provided.
In another resampling circuit, resampling is again performed along the x-axis first, and subsequently followed by resampling along the y-axis. However, rather than accessing the graphics data in the inefficient manner previously described, the resampling circuit includes a plurality of series connected delay buffers having equal lengths that are sufficient to accommodate an entire row of input samples of the source image. The number of delay buffers depends on the number of samples used to interpolate a new sample along the y-axis. Rows of the stored graphics data from the x-axis resampling operation are read into and shifted through the delay buffers, thus allowing for a page or burst mode to be used. Circuitry for performing the interpolation along the y-axis is coupled to the series of delay buffers such that there is a tap at the input of the first delay buffer, and taps at the output of each delay buffer to provide synchronized access to the data used in the y-resampling operation. As the data continues to shift through the delay buffers, the synchronization is maintained. As a result, the graphics data used to obtain the sample history needed for model generation along the y-axis is available to the interpolation circuitry without the access latency present in the previously described conventional resampling process.
In cases where the source image is scaled by {Rx, Ry} to increase image size along the x- and y- axes, respectively, the quantity of intermediate data generated by the resampling circuit along the x-axis is approximately equal to (Rx×source_width), where source_width is the width of the source image. Thus, if generating and evaluating interpolation models along the y-axis requires a history of Ny samples, then a total of (Ny−1) delays, each delay having a length of (Rx_max×source_width_max), where Rx_max is the maximum scaling ratio accommodated by the resampling circuit and source_width_max is the maximum width of a source image resampled by the resampling circuit. As one could imagine, the space occupied by the delays necessary for resampling along the y-axis is considerable. Moreover, the scaling ratio and size of the source image are limited by the widths of the delays in the conventional resampling circuit.
Therefore, there is a need for a flexible graphics resampling system and method that can accommodate a variety of different sized source images and various scaling ratios.